Underpricing

To answer Hypothesis 1, we will have to look at the phenomenon of underpricing during Covid-19 and compare it to the findings previously to Covid-19. To do so, we first conducted an univariate analysis and then a multiple regression model, in line with the model described in Chapter 5. Methodology.

Univariate

As we compared two independent populations to answer hypothesis 1, we conducted a two-sample independent t-test on the following null hypothesis.

H0: The underpricing during Covid-19 is similar to before Covid-19. (Help us answer Hypothesis 1, the Covid-19-effect).

We looked at both underpricing in absolute numbers and the log-transformed excessive underpricing, which was used in our multiple regression. In absolute numbers, we found an average underpricing of 10.3% during Covid-19 and only 6.5% previously to Covid-19. The selection was a total of 73 listings during the Covid-19 period and 145 before. The t-test confirms the difference to be significant.

Looking at the log-transformed excessive underpricing for the same companies, we saw the same picture. The excessive underpricing is significantly higher during Covid-19, yet the difference is smaller. Therefore, given the results from the t-tests, we reject the null hypothesis.

H0: The underpricing is similar on the MTF markets and the leading stock exchanges during Covid-19. (Help us answer Hypothesis 2, the MTF market-effect).

We found some unexpected results when answering our second hypothesis on whether the listings on the Nordic MTF markets are underpriced more than listings on the leading Nordic stock exchanges. We saw that listings on the leading stock exchanges were more underpriced than those on MTF markets for the absolute underpricing. In absolute numbers, the average underpricing on leading stock exchanges summed up to 15%, while the listings on MTF markets were underpriced by just 12%. This said, the difference is not significant. Excessive underpricing follows the same pattern. The MTF market listings are less underpriced, but the difference is not significant.

Multiple regression model

In the following, we will look at a multiple regression model with the insights gotten from this univariate analysis in mind.

Table 13. Multiple Regression on underpricing

The multiple regression model shows five models. Firstly, we look at the effect on being listed at the MTF markets isolated. Secondly, we look at companies listed during Covid-19 in Model 2 before we control the two attributes for each other in Model 3 In Model 4, we look at how

asymmetric information relates to underpricing before we in Model 5 look at all the variables controlled for each other.

In Model 2, we saw the Covid-19 effect isolated. As we see from the results of our regression model, we can see how the excessive underpricing is 3.6% higher during Covid-19 when the variable is isolated and 9.8% higher when controlled for omitted variables, compared to pre-Covid-19, the result is significant at 99% confidence level.

Further, from the multiple regression model, we answered parts of Hypothesis 2 on whether the underpricing during Covid-19 on the MTF markets is greater than the leading exchanges.

In Model 1, one can observe that in general, there is more underpricing at MTF markets, 0.4%, but not during Covid-19, -4.1%. However, none of these findings are significant, and we could not draw any conclusion.

Looking Model 4, we can see the proxies for asymmetric information. FirmAge, FilingToIssue, PrimaryToTotal and Hot have a negative relationship with underpricing.

FirmAge is significantly negative, -0.1% when only looking at proxies for asymmetric information isolated. Adjusting for all variables, the coefficient is insignificant but is still negative, meaning the older the firm is, the less underpricing is there. FilingToIssue is significant adjusted for all control variables. For each marginal day used on preparing the auction, the underpricing is reduced by 0.03%.

From our variable PrimaryToTotal, we can see that IPOs with more primary shares, compared to total shares, experience less underpricing, 3.1%, which is in line with our expectations as it indicates less uncertainty related to these issues. However, it is not statistically significant.

The Hot variable implicates a small negative effect on underpricing. This regression will be further discussed, answering Hypothesis 3 on the effect of asymmetric information on underpricing.

In Model 5, we have, in addition to the abovementioned variables, involved MoneyRaised and the dummy HighTech. We see how the companies classified as HighTech experience more underpricing in the Nordic markets during an IPO, 3.1%, the statistical relationship is insignificant. We did not find any significant statistical relationship between money raised during an IPO and underpricing. If anything, it is weakly positively correlated where a percentage increase in MoneyRaised increases underpricing by 1.4%, meaning the higher the issue size, the higher is the underpricing.

Even though our model left us with some insights, it did not provide a sufficient base to answer our research question. We noticed that the adjusted R is low. Hence it was to a low extent able to predict the variance in the observations. However, the theory deviates from its belief in adjusted R as a good proxy for the model’s explanatory value. We acknowledged the low adjusted R and interpreted the results with caution.

Discussion on our results

Our results build up the recent studies, finding a higher underpricing on listings during the Covid-19-pandemic (Sahac, 2021; Baig & Chen, 2021; Innstrand & Johnsen, 2021). Our findings are in that way further strengthening Ritter's (1991) implications of higher underpricing in windows of opportunities. Furthermore, we find a positive relationship between the MTF market variable and underpricing. This could be interpreted as asymmetric information, given the less comprehensive listing requirements needed to be listed at the multilateral trading facilities, supporting Westerholm's study from 2006. Furthermore, in unison with Ritter's (1991) findings on increased asymmetric information, as young firms emit less public information than mature firms with more extensive financial history.

Our model found some support to asymmetric information being the most influential explanation for the degree of underpricing. The proxy for signaling theory and winner's curse, TransactionToMarketvalue, is insignificantly negative. The negative relationship between the relative size of IPO and underpricing contradicts Keloharju's (1993) findings of an increase in asymmetric information leading to higher underpricing. Further, we found significant numbers for FirmAge to negatively affect ex-ante uncertainty and presumably lead to lower underpricing, yet the coefficient is insignificant when adjusted for omitted variables. The finding that the more mature the firm is, the lower the underpricing is supported by theory (Jenkinson & Ljungqvist, 2001; Rock, 1986).

Looking at the proxy for signal theory, PrimaryToTotal, which Ljungqvuist (2007) found to have a negative relationship with underpricing, is negative in our model. The negative relationship between the portion of issuance of primary shares and underpricing is in one way intuitive. One would expect that a more significant portion of primary shares would signal belief in the company from insiders. External investors should then demand little initial returns; hence the underpricing should be lower. Our results, supporting Ljunqvist (2007), diverge from the results provided by Spiess & Pettway (1997) and Garfinkel (1993), which did not find any relationship between the portion of primary shares issued and underpricing.

The proxy FilingToIssue returns a negative coefficient, indicating that the longer the time between the IPO and IPO announcement, the lower the underpricing. This supports the theory of Ekkayokkaya & Pengniti (2012), who assumed that the longer the time available for analysts to value a company, the less the underpricing would be. In contrast, it contradicts the theory provided by Lee, Taylor, & Walter (1996), who found the underpricing to increase when FilingToIssue increased. The authors explained the effect of extended FilingToIssue signaling lower investor demand. Ekkayokkaya and Pengniti studied Finnish IPOs, while Lee, Taylor & Walter studied U.S. IPOs. One could argue that the limited financial ecosystem in the Nordic markets compared to the U.S, which has interests from all over the world, has led us to the same finding as the Finnish study by Ekkayokkaya and Pengniti.

However, our last proxy for asymmetric information, Hot, shows a negative relationship to underpricing. The rationale behind the proxy was that during hot periods, the limited supply of underwriters would increase their hedge against overpricing, setting the IPO price unnaturally low (Ibbotson & Ritter, 1995; Abrahamsson & De Ridder, 2015). Nevertheless, our finding suggests that there is lower underpricing during hot periods.

After all, we have found some interesting results indicating that the Nordic IPO markets are not necessarily following the same patterns as theorists in the U.S have suggested. Our results support Westerholm’s (2006) findings of lower underpricing in the Nordic markets than in the U.S. Further, our research question surrounding underpricing during Covid-19 shows a strong sign towards higher underpricing during this hot issue market. Lastly, we found divergent results on asymmetric information being as an explanatory factor for underpricing.